Constellation Software 2026 AGM — Part II: On Topicus & Lumine

What the Spin-offs Are Building

---

If you read the first part of this AGM series 🔗 Constellation Software AGM 2026 - Part I, you already know what Mark Miller and the CSU team had to say. But other revealing moments of the event came not from Constellation itself, but from its two spin-offs: Topicus and Lumine.

Robin van Poelje and David Nyland both took the stage and, in their own very different ways, gave investors a clearer window into what these businesses are today; and, more importantly, what they are becoming.

I’ll cover each separately, because there were several important nuggets worth unpacking.

---

Part I: Robin van Poelje & the Topicus DNA

Topicus is not exactly a mini-CSU

One of the most common mistakes I see investors make (I include myself) when analyzing Topicus is treating it as a straightforward Constellation clone. The more I learn about it, the more I feel it isn’t, and van Poelje was explicit about why during the AGM.

Topicus came into the CSU ecosystem with a different operating culture, one built around product development and organic growth, and the decision was made early on to protect that, rather than overwrite it with the typical TSS playbook:

“I always think organic growth [and] acquisitive growth. I like them both... when we did the spinout together with Topicus, they all historically had very strong organic growth... So we didn’t want to saturate and kill that with a typical TSS approach.”

This is worth sitting with for a moment. The typical TSS approach, i.e. focused primarily on acquired growth and operational optimization, would have been the path of least resistance. Van Poelje and his management team chose not to take it. Incentive structures at Topicus are specifically designed to reward organic growth alongside acquired growth. The company is, by design, operating as a true software builder and not just a financial holding company. That distinction matters quite a bit for how you think about Topicus’s long-run potential and its ability to sustain value creation through product iteration.

Geographic expansion

In the Q&A, some shareholders raised the question of Topicus’s recent acquisition in Indonesia, a clear departure from its historically pure-play European focus. Van Poelje confirmed the obvious: Topicus is “slowly but surely moving out of Europe.” But he was careful to contextualize what that actually means in practice.

The geographic expansion is not really a pivot but a slow, step-by-step process. Van Poelje was honest about the operational realities that demand this level of caution. The distance alone (14 hours to Jakarta vs 1-2h to any European region) and the associated cultural and implementation challenges mean that copying-and-pasting the European playbook into Southeast Asia would be a mistake. Success in these markets, as it tends to be in VMS, depends heavily on the quality of local management. You can have the best M&A process in the world and still get burned if you don’t have the right people on the ground.

So, with moves like these, Topicus is clearly trying to expand its TAM beyond Europe.

That brings me to a point raised informally during the AGM that I think is underappreciated. Someone observed that if Topicus had all the opportunities it needed in Europe, it probably wouldn’t be looking at Indonesia and other international markets. I think there’s some truth to that.

The marginal deal in Europe is becoming harder to find at the right price. Increased competition from copycat acquirers and the complexity of navigating multiple regulatory regimes create a double-edged sword: they can act as barriers to entry, but they also make M&A execution more difficult and costly.

Van Poelje knows this. Expanding the hunting ground is the rational response.

One additional data point: van Poelje also mentioned that Topicus already has people on the ground in Asia, but also the US, and Latin America. In other words, the infrastructure for international expansion is already being assembled.

We’ve also seen this in practice. One of Topicus’ most meaningful recent acquisitions was Keypoint Intelligence in the US. As one Substack reader pointed out, the business could be generating around $50M in annual revenue. While I haven’t independently verified that figure, it does reinforce the broader point: Topicus appears to be doing more than simply testing the waters outside its traditional European turf.

The company seems increasingly willing to pursue meaningful opportunities abroad. Whether this becomes a defining part of the next growth chapter remains to be seen, but the early signs are certainly there.

The Asseco Poland experiment

The PEMS strategy generated a fair amount of discussion at the AGM, and I think it’s one of the more interesting strategic experiments within the broader CSU ecosystem. For those who need the context: Topicus took a minority stake in Asseco Poland roughly six months ago, securing three board seats in the process.

The most important thing to understand here is what this relationship is not. It is not a hostile takeover play. It is not a forced integration into the Constellation operating system.

Van Poelje’s framing was clear: the goal is a value-add partnership where Topicus shares best practices and the two entities can generate mutual benefits without destroying what Asseco has already built. “1 plus 1 is more than 2” was how he put it.

Also, the M&A overlap question (Topicus vs Asseco) is worth addressing directly because it came up at the event. Asseco is not (really) part of the Constellation system, which means it runs its own M&A strategy independently. Topicus and Asseco have conflict-of-interest resolutions in place, but Asseco is free to compete for assets on its own terms. Whether this creates friction over time is something I’ll be watching closely. For now, it seems manageable, but the proof will be in how the relationship develops over the next 12–24 months.

On AI and self-disruption

Van Poelje was asked whether Topicus would use AI to disrupt its own legacy businesses from scratch. As a good Dutchman, his answer was rather pragmatic, and I thought it was the right one:

“It’s easily said, I’m going to disrupt my own business... [it’s] easier said than done.”

He noted that if Topicus were to attempt self-disruption, it would require organizing a dedicated team outside the existing business unit, insulated from the cultural inertia of the incumbent. And, he suggested this is best targeted at their weaker businesses, where the delta between what exists today and what a clean-sheet rebuild could achieve is large enough to justify the disruption cost. That’s a thoughtful framework, and it’s the kind of disciplined thinking that I expect from a company like Topicus.

On the AI productivity question more broadly, van Poelje acknowledged what most software operators will tell you: yes, AI is driving meaningful productivity gains in development, and yes, Topicus will capture those gains.

A few other notes from the floor conversations with van Poelje that I think are worth flagging. Full credit goes to X user Raj.brk, who took excellent notes throughout the event while attending in person and had the opportunity to speak one-on-one with several members of the CSU management team.

• On listing Topicus on US exchanges: he showed no interest. Managing two listing fees and two exchange relationships is not something he wants to deal with. This isn’t a particularly bullish or bearish data point, but it does suggest he is focused on running the business rather than on financial engineering.

• On the Scandinavian copycats (Chapters Group and others): van Poelje acknowledged the competitive dynamic bluntly — “They steal our guys.” Talent poaching from the CSU ecosystem is real and ongoing. Whether this creates meaningful long-term pressure on Topicus’s ability to compound is a legitimate question, but I don’t think the answer is yes yet.

• On the pure vibes front when speaking with van Poelje in person: by all accounts, van Poelje came across as extremely confident and well-prepared. He handled difficult questions without flinching. Not a financial metric, but it matters.

---

Part II: David Nyland & the Lumine Advantage

The vertical no one wants to compete in

If Topicus is the more accessible of the two spinoffs from a narrative standpoint, Lumine is the more interesting (and complex) one from a structural moat perspective. CEO D. Nyland’s commentary at the AGM was, in my view, some of the most strategically revealing of the entire event, and a lot of it centered on something that doesn’t get discussed nearly enough: just how difficult and unattractive Lumine’s vertical is for anyone who isn’t already embedded in it.

Lumine operates primarily in telecom and media software, with 94% of its revenue coming from Tier 1 operators. These are not nimble, cloud-native organizations that can flip to a new vendor in six months. They are inherently conservative businesses, heavily regulated, still largely running on bare-metal servers or private clouds, and, in Nyland’s words, “bruised and fatigued” from past technology cycles. They have been sold on transformative technology before, and many of them got burned. That institutional memory runs deep.

This context is essential for understanding everything else Nyland said, particularly around AI.

AI in telecom: a 5-to-10 year story, not a 12-month one

The AI hype cycle has produced no shortage of commentary about how quickly LLMs will disrupt enterprise software. Nyland offered a different view, grounded in the actual behavior of his customers rather than in speculation about what AI is theoretically capable of:

“Agentic AI adoption [in Tier 1 telecom] is going to be over the next 5 to 10 years. They want to see other people do it first. They want to see stuff that’s secure, passes regulatory control.”

To me, this is a realistic statement about how regulated, risk-averse enterprises adopt new technology. Tier 1 telecoms will not be early adopters of anything that touches core processing, and AI is no exception. They want case studies. They want regulatory clarity. They want to see someone else absorb the implementation risk before they take it on.

The interesting flip side of this is what it means for Lumine internally. Since customers are not yet buying AI-enhanced products, Lumine is deploying AI to compress its own development and testing cycles. Nyland specifically mentioned that the massive regression and security testing cycles that previously ran on 3-year timelines are being compressed to 1 year, and potentially down to 6 months.

For a business built around serving deeply complex, regulated customers, this is no trivial operational improvement. It’s a structural cost reduction that could potentially show up in margins over time without requiring any change in customer behavior.

Corporate carve-outs as a competitive moat

Another underappreciated aspect of Lumine’s M&A strategy is its focus on corporate carve-outs. Deals like WideOrbit and Synchronoss, which involve separating a division from a much larger parent company. These transactions are very complex, and that complexity is itself a moat. Most acquirers don’t want to deal with them.

“Corporate carve-outs are extremely complex, especially the larger ones with dysfunctional sellers.”

The phrase “dysfunctional sellers” is doing a lot of work in that sentence. Large corporate parents divesting non-core software assets are often disorganized, slow, and difficult to work with. Integration requires navigating legacy systems, retained shared services, and organizational complexity that most buyers find prohibitive. Lumine has built the institutional capability to manage exactly this kind of transaction, and it builds its management infrastructure a full year ahead of when it is needed.

I’ll add my own observation here: the barriers to entry in corporate carve-outs are not just operational. They are reputational. Sellers need to trust that the acquirer can execute the separation without destroying the asset in the process. Lumine has now done this multiple times at scale. That track record matters in deal sourcing.

The go-to-market moat that AI startups cannot replicate

The third piece of Nyland’s commentary that stood out was his dismissal of the AI startup threat and the reasoning he gave for it.

Getting a new logo in Tier 1 telecom is extremely difficult. These customers want vendor consolidation, not experimentation with unproven technology providers. The sales cycle is long, the evaluation process is rigorous, and the switching costs are big.

So, Nyland’s point was simple: Lumine already has 300 customers globally and the deep trust that comes from years of operating critical infrastructure for them. When Lumine acquires a new company, it inherits that trust immediately and can leverage the existing distribution network to cross-sell:

“We don’t have to go and find them. We don’t have to persuade that we can go and have a meeting.”

An AI startup, no matter how good the product, has to earn its way into a Tier 1 telecom over a multi-year sales cycle. Lumine doesn’t. That is the distribution moat. And in a market characterized by conservative, risk-averse customers who actively want consolidation, distribution is more defensible than product.

The WideOrbit equity playbook

A shareholder asked another interesting question: if Lumine can issue its own stock at ~16x EBITDA to acquire assets at ~12x EBITDA, why isn’t it doing more of these deals? Nyland’s answer was, once again, instructive.

He called WideOrbit a “premium asset” and a “very unique deal,” and reiterated that Lumine’s core thesis remains compounding with cash. But he also left the door open:

“If we trade above intrinsic value, I think there might be an opportunity to use the [equity] in a deal.”

This is worth watching. Lumine has historically been conservative about using its own stock as currency, and for good reason, given that the discipline around capital allocation is part of what makes the CSU ecosystem special. But if the stock continues to trade at a significant premium to intrinsic value, the math on equity-funded acquisitions can become compelling, particularly for large, complex carve-outs where the capital requirement exceeds what cash reserves can comfortably support.

Talent is the binding constraint

The last point from Nyland also deserves attention when thinking about Lumine’s long-run compounding potential. He was clear to point out that you don’t scale this CSU-type model without an active, ongoing talent management process. The key is placing the right people a full year before you actually need them:

“You just don’t scale unless you’ve got an ongoing active talent management process and you’re harvesting talent... we got to throw a lot of world-class fertilizer on that field.”

This is the core competency of the business, and it is the thing that is hardest to replicate. The deal pipeline, the M&A expertise, the carve-out capability… none of it matters if the talent infrastructure is not ahead of the growth curve. The fact that Nyland flagged this as the binding constraint suggests he is thinking about it the right way.

I’ve got to say, I have a real soft spot for Nyland’s leadership style. More on this on my over takeaway below.

---

My Overall Takeaway

What I took away from both sessions is that Topicus and Lumine are trying to building durable, differentiated franchises, but for different reasons and through very different mechanisms.

Topicus is interesting because it is trying to do something that no one in the CSU ecosystem does as good as them: grow organically as well as through acquisitions, while carefully extending its geographic reach in a way that doesn’t break the operating model. They are the precursors of the PEMS strategy and the Asseco experiment is a the first strategic variable worth tracking across the CSU universe. The international expansion feels real, even if it is measured.

Lumine is interesting because the structural moat around its business. The combination of customer conservatism, carve-out complexity, and distribution depth makes it one of the harder VMS businesses to disrupt that I can think of. AI is coming to telecom software eventually, but the timeline is long and Lumine is using the intervening period to reduce its own cost structure while fortifying its customer relationships.

Neither business is without risk. Topicus faces a tightening European deal environment and talent competition from an increasingly crowded copycat ecosystem. Lumine is heavily concentrated in Tier 1 telecom, and its growth depends on finding large, complex assets that meet its return thresholds, a bar that gets harder to clear as the business scales.

But on balance, both van Poelje and Nyland gave me confidence that the management teams running these businesses understand exactly what they own, exactly where the risks lie, and exactly what not to do.

Particularly Nyland.

As I’ve mentioned before, I really like his style. He comes across as a natural, unapologetic leader who is deeply aligned with the CSU culture built over decades by Mark Leonard. There is a level of clarity, discipline, and long-term thinking that feels authentic rather than rehearsed.

He just gets it.

In this market, that counts for quite a lot.

Thanks for following along,

—Nikotes

---

Unlock Premium Content – For just $0.39/day ($12/month) or $0.27/day ($100/year)!

🔗 Portfolio Corner - 🗓️ Monthly Updates (Last Update: 3-May-2026)